Fix XX reduction for XX Codes with HLA- prefix (#165)

* Fix XX reduction for XX Codes with HLA- prefix

* Bump version: 0.7.6 → 0.7.7

Author: pbashyal-nmdp

pyard/__init__.py

@@ -24,5 +24,5 @@
 from .pyard import ARD
 
 __author__ = """NMDP Bioinformatics"""
-__version__ = '0.7.6'
+__version__ = '0.7.7'
 

pyard/pyard.py

@@ -233,7 +233,7 @@ def sorted_unique_gl(self, gls: List[str], delim: str) -> str:
         if delim == '~':
             # No need to sort
             return delim.join(gls)
-        
+
         if delim == "+":
             # No need to make unique. eg. homozygous cases are valid for SLUGs
             return delim.join(sorted(gls, key=functools.cmp_to_key(smart_sort_comparator)))
@@ -296,8 +296,16 @@ def redux_gl(self, glstring: str, redux_type: VALID_REDUCTION_TYPES) -> str:
             raise InvalidTypingError(f"{glstring} is not a valid V2 or Serology typing.")
 
         # Handle XX codes
-        if self._config["reduce_XX"] and self.is_XX(glstring, loc_antigen, code):
-            return self.redux_gl("/".join(self.xx_codes[loc_antigen]), redux_type)
+        if self._config["reduce_XX"]:
+            is_hla_prefix = HLA_regex.search(loc_antigen)
+            if is_hla_prefix:
+                loc_antigen = loc_antigen.split('-')[1]
+            if self.is_XX(glstring, loc_antigen, code):
+                if is_hla_prefix:
+                    reduced_alleles = self.redux_gl("/".join(self.xx_codes[loc_antigen]), redux_type)
+                    return "/".join(["HLA-" + a for a in reduced_alleles.split("/")])
+                else:
+                    return self.redux_gl("/".join(self.xx_codes[loc_antigen]), redux_type)
 
         # Handle MAC
         if self._config["reduce_MAC"] and self.is_mac(glstring):
@@ -523,7 +531,7 @@ def isvalid(self, allele: str) -> bool:
                 allele = allele[:-1]
                 if self._is_valid_allele(allele):
                     return True
-                else: 
+                else:
                     allele = get_2field_allele(allele)
                     if self._is_valid_allele(allele):
                         return True

setup.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.7.6
+current_version = 0.7.7
 commit = True
 tag = True
 

setup.py

@@ -42,7 +42,7 @@
 
 setup(
     name='py-ard',
-    version='0.7.6',
+    version='0.7.7',
     description="ARD reduction for HLA with Python",
     long_description=readme + '\n\n' + history,
     long_description_content_type="text/markdown",

tests/test_pyard.py

@@ -97,6 +97,13 @@ def test_xx_code(self):
         gl = self.ard.redux_gl('B*40:XX', 'G')
         self.assertEqual(gl, expanded_string)
 
+    def test_xx_code_with_prefix(self):
+        expanded_string = """
+        HLA-B*40:01:01G/HLA-B*40:01:03G/HLA-B*40:02:01G/HLA-B*40:03:01G/HLA-B*40:04:01G/HLA-B*40:05:01G/HLA-B*40:06:01G/HLA-B*40:07/HLA-B*40:08/HLA-B*40:09/HLA-B*40:10:01G/HLA-B*40:11:01G/HLA-B*40:12/HLA-B*40:13/HLA-B*40:14/HLA-B*40:15/HLA-B*40:16:01G/HLA-B*40:18/HLA-B*40:19/HLA-B*40:20:01G/HLA-B*40:21/HLA-B*40:22N/HLA-B*40:23/HLA-B*40:24/HLA-B*40:25/HLA-B*40:26/HLA-B*40:27/HLA-B*40:28/HLA-B*40:29/HLA-B*40:30/HLA-B*40:31/HLA-B*40:32/HLA-B*40:33/HLA-B*40:34/HLA-B*40:35/HLA-B*40:36/HLA-B*40:37/HLA-B*40:38/HLA-B*40:39/HLA-B*40:40:01G/HLA-B*40:42/HLA-B*40:43/HLA-B*40:44/HLA-B*40:45/HLA-B*40:46/HLA-B*40:47/HLA-B*40:48/HLA-B*40:49/HLA-B*40:50:01G/HLA-B*40:51/HLA-B*40:52/HLA-B*40:53/HLA-B*40:54/HLA-B*40:57/HLA-B*40:58/HLA-B*40:59/HLA-B*40:60/HLA-B*40:61/HLA-B*40:62/HLA-B*40:63/HLA-B*40:64:01G/HLA-B*40:65/HLA-B*40:66/HLA-B*40:67/HLA-B*40:68/HLA-B*40:69/HLA-B*40:70/HLA-B*40:71/HLA-B*40:72/HLA-B*40:73/HLA-B*40:74/HLA-B*40:75/HLA-B*40:76/HLA-B*40:77/HLA-B*40:78/HLA-B*40:79/HLA-B*40:80/HLA-B*40:81/HLA-B*40:82/HLA-B*40:83/HLA-B*40:84/HLA-B*40:85/HLA-B*40:86/HLA-B*40:87/HLA-B*40:88/HLA-B*40:89/HLA-B*40:90/HLA-B*40:91/HLA-B*40:92/HLA-B*40:93/HLA-B*40:94/HLA-B*40:95/HLA-B*40:96/HLA-B*40:98/HLA-B*40:99/HLA-B*40:100/HLA-B*40:101/HLA-B*40:102/HLA-B*40:103/HLA-B*40:104/HLA-B*40:105/HLA-B*40:106/HLA-B*40:107/HLA-B*40:108/HLA-B*40:109/HLA-B*40:110/HLA-B*40:111/HLA-B*40:112/HLA-B*40:113/HLA-B*40:114:01G/HLA-B*40:115/HLA-B*40:116/HLA-B*40:117/HLA-B*40:118N/HLA-B*40:119/HLA-B*40:120/HLA-B*40:121/HLA-B*40:122/HLA-B*40:123/HLA-B*40:124/HLA-B*40:125/HLA-B*40:126/HLA-B*40:127/HLA-B*40:128/HLA-B*40:129/HLA-B*40:130/HLA-B*40:131/HLA-B*40:132/HLA-B*40:133Q/HLA-B*40:134/HLA-B*40:135/HLA-B*40:136/HLA-B*40:137/HLA-B*40:138/HLA-B*40:139/HLA-B*40:140/HLA-B*40:142N/HLA-B*40:143/HLA-B*40:145/HLA-B*40:146/HLA-B*40:147/HLA-B*40:148/HLA-B*40:149/HLA-B*40:152/HLA-B*40:153/HLA-B*40:154/HLA-B*40:155:01G/HLA-B*40:156/HLA-B*40:157/HLA-B*40:158/HLA-B*40:159/HLA-B*40:160/HLA-B*40:161/HLA-B*40:162/HLA-B*40:163/HLA-B*40:164/HLA-B*40:165/HLA-B*40:166/HLA-B*40:167/HLA-B*40:168/HLA-B*40:169/HLA-B*40:170/HLA-B*40:171/HLA-B*40:172/HLA-B*40:173/HLA-B*40:174/HLA-B*40:175/HLA-B*40:177/HLA-B*40:178/HLA-B*40:180/HLA-B*40:181/HLA-B*40:182/HLA-B*40:183/HLA-B*40:184/HLA-B*40:185/HLA-B*40:186/HLA-B*40:187/HLA-B*40:188/HLA-B*40:189/HLA-B*40:190/HLA-B*40:191/HLA-B*40:192/HLA-B*40:193/HLA-B*40:194/HLA-B*40:195/HLA-B*40:196/HLA-B*40:197/HLA-B*40:198/HLA-B*40:199/HLA-B*40:200/HLA-B*40:201/HLA-B*40:202/HLA-B*40:203/HLA-B*40:204/HLA-B*40:205/HLA-B*40:206/HLA-B*40:207/HLA-B*40:208/HLA-B*40:209/HLA-B*40:210/HLA-B*40:211/HLA-B*40:212/HLA-B*40:213:01G/HLA-B*40:214/HLA-B*40:215/HLA-B*40:216N/HLA-B*40:217/HLA-B*40:218/HLA-B*40:219/HLA-B*40:220/HLA-B*40:222/HLA-B*40:223/HLA-B*40:224/HLA-B*40:225/HLA-B*40:226/HLA-B*40:227/HLA-B*40:228/HLA-B*40:230/HLA-B*40:231/HLA-B*40:232/HLA-B*40:233/HLA-B*40:234/HLA-B*40:235/HLA-B*40:237/HLA-B*40:238/HLA-B*40:239/HLA-B*40:240/HLA-B*40:242/HLA-B*40:243/HLA-B*40:244/HLA-B*40:245/HLA-B*40:246/HLA-B*40:248/HLA-B*40:249/HLA-B*40:250/HLA-B*40:251/HLA-B*40:252/HLA-B*40:253/HLA-B*40:254/HLA-B*40:255/HLA-B*40:256N/HLA-B*40:257/HLA-B*40:258/HLA-B*40:259/HLA-B*40:260/HLA-B*40:261/HLA-B*40:262/HLA-B*40:263N/HLA-B*40:265N/HLA-B*40:266/HLA-B*40:268/HLA-B*40:269/HLA-B*40:270/HLA-B*40:271/HLA-B*40:273/HLA-B*40:274/HLA-B*40:275/HLA-B*40:276/HLA-B*40:277/HLA-B*40:279/HLA-B*40:280/HLA-B*40:281/HLA-B*40:282/HLA-B*40:283/HLA-B*40:284/HLA-B*40:285/HLA-B*40:286N/HLA-B*40:287/HLA-B*40:288/HLA-B*40:289/HLA-B*40:290/HLA-B*40:291N/HLA-B*40:292/HLA-B*40:293/HLA-B*40:294/HLA-B*40:295/HLA-B*40:296/HLA-B*40:297/HLA-B*40:298/HLA-B*40:300/HLA-B*40:302/HLA-B*40:304/HLA-B*40:305/HLA-B*40:306/HLA-B*40:307/HLA-B*40:308/HLA-B*40:309/HLA-B*40:310/HLA-B*40:311/HLA-B*40:312/HLA-B*40:313/HLA-B*40:314/HLA-B*40:315/HLA-B*40:316/HLA-B*40:317/HLA-B*40:318/HLA-B*40:319/HLA-B*40:320/HLA-B*40:321/HLA-B*40:322/HLA-B*40:323/HLA-B*40:324/HLA-B*40:325/HLA-B*40:326/HLA-B*40:327/HLA-B*40:328/HLA-B*40:330/HLA-B*40:331/HLA-B*40:332/HLA-B*40:333/HLA-B*40:334/HLA-B*40:335/HLA-B*40:336/HLA-B*40:337N/HLA-B*40:339/HLA-B*40:340/HLA-B*40:341/HLA-B*40:342/HLA-B*40:343/HLA-B*40:344/HLA-B*40:345N/HLA-B*40:346/HLA-B*40:347/HLA-B*40:348/HLA-B*40:349/HLA-B*40:350/HLA-B*40:351/HLA-B*40:352/HLA-B*40:354/HLA-B*40:355/HLA-B*40:357/HLA-B*40:358/HLA-B*40:359/HLA-B*40:360/HLA-B*40:361N/HLA-B*40:362/HLA-B*40:363/HLA-B*40:364/HLA-B*40:365/HLA-B*40:366/HLA-B*40:367/HLA-B*40:368/HLA-B*40:369/HLA-B*40:370/HLA-B*40:371/HLA-B*40:372N/HLA-B*40:373/HLA-B*40:374/HLA-B*40:375/HLA-B*40:376/HLA-B*40:377/HLA-B*40:378/HLA-B*40:380/HLA-B*40:381/HLA-B*40:382/HLA-B*40:385/HLA-B*40:388/HLA-B*40:389/HLA-B*40:390/HLA-B*40:391/HLA-B*40:392/HLA-B*40:393/HLA-B*40:394/HLA-B*40:396/HLA-B*40:397/HLA-B*40:398/HLA-B*40:399N/HLA-B*40:400/HLA-B*40:401/HLA-B*40:402/HLA-B*40:403/HLA-B*40:404/HLA-B*40:407/HLA-B*40:408/HLA-B*40:409/HLA-B*40:410/HLA-B*40:411/HLA-B*40:412/HLA-B*40:413/HLA-B*40:414/HLA-B*40:415/HLA-B*40:420/HLA-B*40:421Q/HLA-B*40:422/HLA-B*40:423/HLA-B*40:424/HLA-B*40:426N/HLA-B*40:428N/HLA-B*40:429/HLA-B*40:430/HLA-B*40:432/HLA-B*40:433/HLA-B*40:434/HLA-B*40:436/HLA-B*40:437/HLA-B*40:438N/HLA-B*40:441/HLA-B*40:445/HLA-B*40:447/HLA-B*40:448/HLA-B*40:449/HLA-B*40:451/HLA-B*40:452/HLA-B*40:454/HLA-B*40:457/HLA-B*40:458/HLA-B*40:459/HLA-B*40:460/HLA-B*40:461/HLA-B*40:462/HLA-B*40:463/HLA-B*40:465/HLA-B*40:466/HLA-B*40:467/HLA-B*40:468/HLA-B*40:469/HLA-B*40:470/HLA-B*40:471/HLA-B*40:472/HLA-B*40:477/HLA-B*40:478/HLA-B*40:479/HLA-B*40:481N/HLA-B*40:482
+        """.strip()
+        gl = self.ard.redux_gl('HLA-B*40:XX', 'G')
+        self.assertEqual(expanded_string, gl)
+
     def test_expand_mac(self):
         mac_ab_expanded = ['A*01:01', 'A*01:02']
         self.assertEqual(self.ard.expand_mac('A*01:AB'), mac_ab_expanded)